The exponent of a polarizing matrix constructed from the Kronecker product (Q2439734)

The paper is a continuation of [\textit{S. B. Korada, E. Şaşoglu} and \textit{R. Urbanke}, ``Polar codes: characterization of exponent, bounds, and constructions'', IEEE Trans. Inf. Theory 56, No. 12, 6253--6264 (2010)], defining a new result concerning polar codes: in a symmetric binary-input discrete memoryless channel, the partial distances (and therefore the exponent) of polarizing matrix \(A\otimes B\) -- where \(\otimes\) is the Kronecker product -- can be effectively obtained from partial distances (exponents) of matrices \(A\) and \(B\). These two results are presented in Theorem 7 and respectively Corollary 8; the rest of the paper is mainly dedicated to the proof of these assertions. The paper is interesting, rigorously written, valuable by its result -- which can be extended to a non-binary case, and by its potential applications in all codes based by Kronecker product. As a small critical remark, there are some weaknesses in writing the paper. For example, the main notion - polarizing matrix - is not defined, and the reader has to look for it in the references. Also, for the same term \(E(G)\) two names are used: \textit{rate of polarization} and \textit{exponent}; this seems to be an ambiguity inherited from the paper cited in the beginning.